9 réflexions au sujet de “G : Test de vitesse Nougat Android vs vs Oreo Pie vs vs Android Q”

  1. I’d like to see a reproducibility study of these benchmarks. Run it 10 times, of course not in a row; in a way to avoid thermal effects. How much variation is there?

    Because N is behind until it hits the SQLite portion, and that’s where it gains against 9 then it’s about even between 7 and 8, then 7 slowly creeps ahead. I’d love to see the 10x reproduction of this, I’m betting it’ll just vary a lot.

  2. I’m not sure this really proves anything. It’s not taking into account improvements made in newer versions of android to reduce jank which can be hard to benchmark if you’re not looking at the right metrics (pure speed isn’t always one of them).

    Even though Nougat may finish first Pie and Q may feel smoother to the user because of improvements made to reduce jank such as reducing the GC pause time. Some apps may not benefit as much from the ART improvements either depending on their code makeup.

    While the above focuses on ART there can also be a number of other OS changes between major android versions such as security features introduced and possible changes in scheduling.

    tl;dr looks like click bait and contributes nothing new

  3. Is this not incredibly obvious though? Each new version is designed for more powerful hardware, and therefore should be slower (but overall faster on the newer hardware).

    Imagine running Windows 3.1 on today’s hardware. Assuming you had all the drivers and stuff supported, it should never drop a frame. Now try running Windows 10 on even an early 2000s machine.

Les commentaires sont fermés.